Agent Browser
Text-based browser automation CLI for AI agents
Real-browser testing for agents without burning tokens
A command-line interface for common Playwright actions from Microsoft that lets an agent drive a real browser, click through an app and record repeatable tests. It suits developers who need proof that a build actually works before handing it to a client.
Instead of loading a full MCP page dump into context, the agent issues CLI commands to open a browser, inspect selectors, take screenshots and generate Playwright code it can rerun. That keeps token usage low and the loop fast while still verifying real user flows.
I Tried 100+ Github Repos for Claude. These 6 Are the Best
Text-based browser automation CLI for AI agents
Token-efficient browser control for AI agents via page structure
Karpathy-inspired skills for planning, simplicity and surgical edits
Open-source AI coding agent you can customize and self-host
Agentic coding tool in the terminal that understands your codebase
Open-source terminal coding agent you can fork and rebrand